rudachi 1.1.1 → 1.4.0

checksums.yaml

@@ -1,7 +1,7 @@
 ---
 SHA256:
-  metadata.gz: 1cd976cea8211f4f452c2cd20a3fb286828d449aa346ea54458e98ecf47819aa
-  data.tar.gz: 7b8b886b6dc1467d33fdd44878ce1a490c13c7aacd6a40bd80e68ddcea631801
+  metadata.gz: 4b76fd7addfd5f3bd28c0a037a2b3953ddb05a1f83af90aa154e64f54bdd4487
+  data.tar.gz: 57475d77337ee0e058672827ea83ccca6d2713543431deb8625ae6cce4e88abf
 SHA512:
-  metadata.gz: 6c21aaea246963a585ceee7800f13013797ae16f10386c933e01d3f1709ed96a5a7bcfd0d5f0e9e0a43ba17b9be2a7d2db0ce9a82ec446673ff2cf3b48a016fa
-  data.tar.gz: c06e0239aa4c8bae28f9a4b60d00beb66aea07a7d85760126a2e875e539a46a6716aace54f62f3f193edee31c8e8300c34bf1b2be8908b4acc2285d366bc1b7b
+  metadata.gz: a07a1ac681d0999c103ca0f27c12190a8fa95de9ab7a5613d7d6412709ce55b4eb4492c471a1fb10bf5d238b205e51b281b3cd5a6f54237577463a3a8563cbd2
+  data.tar.gz: 75c332d969191be48746969cd103716a3a2c02b7bf1222eaa11b293165635466aeddf93deaed367aba86dd2307fc3df8896eb81b255600fe87e8fa06a599b886

data/README.md

@@ -1,23 +1,29 @@
 # Rudachi
-[Sudachi](https://github.com/WorksApplications/Sudachi) wrapper Gem for JRuby.
+JRuby wrapper for [Sudachi](https://github.com/WorksApplications/Sudachi).
 
-#### For Text
+#### Text
 ```rb
 Rudachi::TextParser.parse('東京都へ行く')
 => "東京都\t名詞,固有名詞,地名,一般,*,*\t東京都\nへ\t助詞,格助詞,*,*,*,*\tへ\n行く\t動詞,非自立可能,*,*,五段-カ行,終止形-一般\t行く\nEOS\n"
 ```
 
-#### For File
+#### File
 ```rb
-File.open('sample.txt', 'w') { |f| f << '東京都へ行く' }
-Rudachi::FileParser.parse('sample.txt')
+File.open('input.txt', 'w') { |f| f << '東京都へ行く' }
+Rudachi::FileParser.parse('input.txt')
+=> "東京都\t名詞,固有名詞,地名,一般,*,*\t東京都\nへ\t助詞,格助詞,*,*,*,*\tへ\n行く\t動詞,非自立可能,*,*,五段-カ行,終止形-一般\t行く\nEOS\n"
+```
+
+#### IO
+```rb
+Rudachi::StreamParser.parse(StringIO.new('東京都へ行く'))
 => "東京都\t名詞,固有名詞,地名,一般,*,*\t東京都\nへ\t助詞,格助詞,*,*,*,*\tへ\n行く\t動詞,非自立可能,*,*,五段-カ行,終止形-一般\t行く\nEOS\n"
 ```
 
 #### With [some options](https://github.com/WorksApplications/Sudachi#options)
 ```rb
-Rudachi::TextParser.new(o: 'result.txt', m: 'A').parse('東京都へ行く')
-File.read('result.txt')
+Rudachi::TextParser.new(o: 'output.txt', m: 'A').parse('東京都へ行く')
+File.read('output.txt')
 => "東京\t名詞,固有名詞,地名,一般,*,*\t東京\n都\t名詞,普通名詞,一般,*,*,*\t都\nへ\t助詞,格助詞,*,*,*,*\tへ\n行く\t動詞,非自立可能,*,*,五段-カ行,終止形-一般\t行く\nEOS\n"
 ```
 
@@ -26,6 +32,8 @@ File.read('result.txt')
 - [JRuby](https://github.com/jruby/jruby) 9.1.3.0 or later
 - [Sudachi](https://github.com/WorksApplications/Sudachi)
 
+For Ruby, please check [rudachi-rb](https://github.com/SongCastle/rudachi-rb).
+
 ## Installation
 
 1. Install JAR and dictionary of Sudachi ([Details](https://github.com/WorksApplications/Sudachi/blob/develop/docs/tutorial.md#linux-%E3%81%AE%E5%A0%B4%E5%90%88))
@@ -54,7 +62,7 @@ LEGAL  LICENSE-2.0.txt	system_core.dic
 gem 'rudachi'
 ```
 
-Then run `bundle install` .
+Then run `bundle install`.
 
 3. Initialize Rudachi
 

data/lib/rudachi/configurable.rb

@@ -11,7 +11,7 @@ module Rudachi
         def self.#{name}; @@#{name}; end
         def self.#{name}=(val); @@#{name} = #{klass}.new(val); end
       EOS
-      public_send("#{name}=", default)
+      send("#{name}=", default)
     end
   end
 end

data/lib/rudachi/dependencies.rb

@@ -1,19 +1,38 @@
 require 'java'
+require 'rudachi/lazy_load'
+require Rudachi.jar_path
+
 java_import 'java.lang.System'
 java_import 'java.io.PrintStream'
 java_import 'java.io.ByteArrayInputStream'
 java_import 'java.io.ByteArrayOutputStream'
 java_import 'java.nio.charset.StandardCharsets'
-
-require Rudachi.jar_path
 java_import 'com.worksap.nlp.sudachi.SudachiCommandLine'
 
-module Java
-  String                = JavaLang::String
-  System                = JavaLang::System
-  ByteArrayInputStream  = JavaIo::ByteArrayInputStream
-  ByteArrayOutputStream = JavaIo::ByteArrayOutputStream
-  PrintStream           = JavaIo::PrintStream
-  UTF_8                 = JavaNioCharset::StandardCharsets::UTF_8
-  SudachiCommandLine    = ComWorksapNlpSudachi::SudachiCommandLine
+module Rudachi
+  module Java
+    String                = ::Java::JavaLang::String
+    System                = ::Java::JavaLang::System
+    PrintStream           = ::Java::JavaIo::PrintStream
+    ByteArrayInputStream  = ::Java::JavaIo::ByteArrayInputStream
+    ByteArrayOutputStream = ::Java::JavaIo::ByteArrayOutputStream
+    UTF_8                 = ::Java::JavaNioCharset::StandardCharsets::UTF_8
+    SudachiCommandLine    = ::Java::ComWorksapNlpSudachi::SudachiCommandLine
+  end
+
+  module StreamProcessor
+    class InvalidInclusion < StandardError; end
+
+    def self.included(base)
+      raise InvalidInclusion unless base.ancestors.include?(TextParser)
+    end
+
+    private
+
+    def input_stream(io)
+      io.to_inputstream
+    end
+
+    LazyLoad.run_load_hooks(:stream_processor, self)
+  end
 end

data/lib/rudachi/file_parser.rb

@@ -8,31 +8,34 @@ module Rudachi
     end
 
     def initialize(**opts)
-      Rudachi.load!
-
-      @output = Java::ByteArrayOutputStream.new
-      @opts   = Option.new(opts)
+      Rudachi::Loader.load!
+      @opts = Option.new(**opts)
     end
 
     def parse(path)
-      take_stdout do
-        Java::SudachiCommandLine.main(
-          Option.cmds(@opts).push(Java::String.new(path))
-        )
+      output_stream do |output|
+        take_stdout(output) do
+          Java::SudachiCommandLine.main(
+            Option.cmds(@opts).push(path)
+          )
+        end
       end
-      @output.toString
     end
 
     private
 
-    def take_stdout
+    def take_stdout(output)
       stdout = Java::System.out
-      stream = Java::PrintStream.new(@output)
-      Java::System.setOut(stream)
 
+      Java::System.setOut(output)
       yield
-
       Java::System.setOut(stdout)
     end
+
+    def output_stream
+      Java::ByteArrayOutputStream.new.tap do |output|
+        yield Java::PrintStream.new(output)
+      end.toString
+    end
   end
 end

data/lib/rudachi/lazy_load.rb

@@ -0,0 +1,18 @@
+module Rudachi
+  module LazyLoad
+    @@hooks = {}
+
+    class << self
+      def on_load(name, &block)
+        @@hooks[name] ||= []
+        @@hooks[name] << block
+      end
+
+      def run_load_hooks(name, mod)
+        @@hooks[name]&.each do |hook|
+          hook.call(mod)
+        end
+      end
+    end
+  end
+end

data/lib/rudachi/loader.rb

@@ -1,16 +1,18 @@
 module Rudachi
-  class UnavailableError < StandardError; end;
+  module Loader
+    class UnavailableError < StandardError; end;
 
-  class << self
-    def load!
-      raise UnavailableError, 'jruby_required' unless jruby?
-      require 'rudachi/dependencies'
-    end
+    class << self
+      def load!
+        raise UnavailableError, 'jruby_required' unless jruby?
+        require 'rudachi/dependencies'
+      end
 
-    private
+      private
 
-    def jruby?
-      RUBY_PLATFORM == 'java'
+      def jruby?
+        RUBY_PLATFORM == 'java'
+      end
     end
   end
 end

data/lib/rudachi/option/boolean_option.rb

@@ -7,8 +7,8 @@ module Rudachi
       end
 
       def __getobj__; @value; end
-      def enable?; @value; end
-      def with_arg?; false; end
+      def enable?   ; @value; end
+      def with_arg? ; false ; end
     end
   end
 end

data/lib/rudachi/option/string_option.rb

@@ -6,9 +6,9 @@ module Rudachi
         @value = str
       end
 
-      def __getobj__; @value; end
-      def enable?; !!@value; end
-      def with_arg?; true; end
+      def __getobj__; @value  ; end
+      def enable?   ; !!@value; end
+      def with_arg? ; true    ; end
     end
   end
 end

data/lib/rudachi/stream_parser.rb

@@ -0,0 +1,10 @@
+require 'rudachi/lazy_load'
+require 'rudachi/text_parser'
+
+module Rudachi
+  class StreamParser < TextParser
+    LazyLoad.on_load(:stream_processor) do |mod|
+      include mod
+    end
+  end
+end

data/lib/rudachi/text_parser.rb

@@ -3,25 +3,29 @@ require 'rudachi/file_parser'
 module Rudachi
   class TextParser < FileParser
     def parse(text)
-      @input = Java::String.new(text)
-      take_stdin do
-        take_stdout do
-          Java::SudachiCommandLine.main(Option.cmds(@opts))
+      output_stream do |output|
+        take_stdin(input_stream(text)) do
+          take_stdout(output) do
+            Java::SudachiCommandLine.main(Option.cmds(@opts))
+          end
         end
       end
-      @output.toString
     end
 
     private
 
-    def take_stdin
+    def take_stdin(input)
       stdin = Java::System.in
-      stream = Java::ByteArrayInputStream.new(@input.getBytes(Java::UTF_8))
-      Java::System.setIn(stream)
 
+      Java::System.setIn(input)
       yield
-
       Java::System.setIn(stdin)
     end
+
+    def input_stream(text)
+      Java::ByteArrayInputStream.new(
+        Java::String.new(text).getBytes(Java::UTF_8)
+      )
+    end
   end
 end

data/lib/rudachi/version.rb

@@ -1,3 +1,3 @@
 module Rudachi
-  VERSION = '1.1.1'
+  VERSION = '1.4.0'
 end

data/lib/rudachi.rb

@@ -2,3 +2,4 @@ require 'rudachi/config'
 require 'rudachi/option/config'
 require 'rudachi/file_parser'
 require 'rudachi/text_parser'
+require 'rudachi/stream_parser'

metadata

@@ -1,16 +1,16 @@
 --- !ruby/object:Gem::Specification
 name: rudachi
 version: !ruby/object:Gem::Version
-  version: 1.1.1
+  version: 1.4.0
 platform: ruby
 authors:
 - SongCastle
 autorequire: 
 bindir: bin
 cert_chain: []
-date: 2022-04-10 00:00:00.000000000 Z
+date: 2022-04-30 00:00:00.000000000 Z
 dependencies: []
-description: Sudachi wrapper for JRuby.
+description: A JRuby wrapper for Sudachi.
 email: "-"
 executables: []
 extensions: []
@@ -22,10 +22,12 @@ files:
 - lib/rudachi/configurable.rb
 - lib/rudachi/dependencies.rb
 - lib/rudachi/file_parser.rb
+- lib/rudachi/lazy_load.rb
 - lib/rudachi/loader.rb
 - lib/rudachi/option/boolean_option.rb
 - lib/rudachi/option/config.rb
 - lib/rudachi/option/string_option.rb
+- lib/rudachi/stream_parser.rb
 - lib/rudachi/text_parser.rb
 - lib/rudachi/version.rb
 homepage: https://github.com/SongCastle/rudachi
@@ -50,5 +52,5 @@ requirements: []
 rubygems_version: 3.0.3
 signing_key: 
 specification_version: 4
-summary: Sudachi wrapper for JRuby
+summary: A JRuby wrapper for Sudachi
 test_files: []